Find Patterns and Anomalies with Data Insights in Autonomous AI Database

Use Data Insights in Data Studio to discover patterns and potential anomalies in data stored in Oracle Autonomous AI Database. Data Insights can generate insights for a table or for an Analytic View, then display the results as charts in the Data Insights dashboard. The tool highlights data points where the actual value of a measure is significantly higher or lower than the expected value for combinations of hierarchy values or table column values.

Use this Autonomous AI Database topic as the entry point for understanding when and how to use Data Insights with Autonomous AI Database. See Data Studio Documentation for complete reference.

When to use Data Insights

Use Data Insights when you want to quickly find unexpected patterns in database data without manually writing analysis queries.

For example, use Data Insights to:

• Find measure values that are unexpectedly high or low.

• Compare actual values against expected values.

• Investigate unusual combinations of dimension or hierarchy values.

• Explore a table before creating reports or dashboards.

• Analyze an Analytic View that was created with Data Analysis.

• Review previous insight searches and reports. 

Data Insights is especially useful during exploratory analysis. It helps identify patterns that may deserve further review, but the results should be interpreted in the business context of the data.

How Data Insights works with Autonomous AI Database

Data Insights runs from Database Actions for Autonomous AI Database. Database Actions is bundled with each Autonomous AI Database instance and includes the Data Studio tools: Data Load, Catalog, Data Insights, Data Transforms, and Data Analysis. See Connect with Built-In Oracle Database Actions for guidance.

You can generate insights for either:

Source	Use it when
Table	You want to analyze data directly from a database table.
Analytic View	You want insights based on business-model metadata, hierarchies, and measures.

Insights generated from an Analytic View can be more useful than insights generated directly from a table because an Analytic View includes additional metadata about dimensions, hierarchies, and measures. See The Data Insights Page for more information.

For non-ADMIN users, Data Studio tools, including Data Insights, require the DWROLE database role. If the Data Insights card is not visible in Database Actions, the database user may not have the required role.

Before you begin

Before using Data Insights:

• Verify that you can access Database Actions for the Autonomous AI Database instance.

• Verify that your database user has the required Data Studio role, such as DWROLE, unless you are using the ADMIN user.

• Make sure the table or Analytic View that you want to analyze is available in your schema or in a schema you can access.

• For richer insight results, consider creating an Analytic View first with the Data Analysis tool. Data Analysis lets you create Analytic Views with multidimensional metadata, hierarchies, and measures. See The Data Analysis Tool for more details.

Open Data Insights

To open Data Insights:

• Open Database Actions for your Autonomous AI Database.

• Select the Data Studio tab.

• Select Data Insights.

You can also open Data Insights from the Database Actions selector menu under Data Tools. 

Generate insights

To generate insights:

• Select a schema.

• Select an Analytic View or table.

• Select the column that contains the measure or data you want to analyze.

• Click Search.

Data Insights submits the request, shows progress while the search runs, and then displays the results in the Data Insights dashboard. 

What Data Insights shows

The Data Insights dashboard displays insight results as bar charts. The charts show actual values, expected values, and the data points with the largest differences between actual and expected values. You can open a chart to see more detail, including the dimension value, a short textual analysis, and the actual and expected values for the selected result. 

You can also:

Task	Description
Refresh results	Refresh the dashboard manually after generating insights.
Enable auto refresh	Let the dashboard refresh automatically while insight generation is in progress.
View errors	Open error details if insight generation fails.
Open chart details	Select a chart to view a larger chart and supporting explanation.
View recent searches	Reopen previous insight searches from the Recent Searches panel.

Notes for Data Insights

• Data Insights identifies potential anomalies. A highlighted result means the value differs from the expected value calculated by the tool; it does not automatically mean that the data is wrong.
• For Analytic Views, Data Insights tends to use higher hierarchy levels because differences at those levels are usually larger in absolute value. The comparison is based on absolute difference, not percentage difference. 
• For tables, Data Insights categorizes columns as dimensions or measures based on data type and cardinality. For example, a VARCHAR2 column is treated as a dimension, while a NUMBER column may be treated as either a dimension or a measure depending on the data.